Merge pull request #208 from tostegroo/improvement/sub-algorithm

New sub/person substitution algorithm.

src/brain.coffee

@@ -987,7 +987,6 @@ class Brain
   # Run substitutions against a message. `type` is either "sub" or "person" for
   # the type of substitution to run.
   substitute: (msg, type) ->
-    result = ""
 
     # Safety checking.
     if not @master._sorted[type]
@@ -997,40 +996,65 @@ class Brain
     # Get the substitutions map.
     subs = if type is "sub" then @master._sub else @master._person
 
-    # Make placeholders each time we substitute something.
-    ph = []
-    pi = 0
-
-    for pattern in @master._sorted[type]
-      result = subs[pattern]
-      qm     = utils.quotemeta pattern
+    # Get the max number of words in sub/person to minimize interations
+    maxwords = if type is "sub" then @master._submax else @master._personmax
 
-      # Make a placeholder.
-      ph.push result
-      placeholder = "\x00<#{pi}>\x00"
-      pi++
+    result = ""
 
-      # Run substitutions.
-      msg = msg.replace(new RegExp("^#{qm}$", "g"),           placeholder)
-      msg = msg.replace(new RegExp("^#{qm}([^a-zA-Z0-9_>]+)", "g"),      "#{placeholder}$1")
-      msg = msg.replace(new RegExp("([^a-zA-Z0-9_<]+)#{qm}([^a-zA-Z0-9_>]+)", "g"), "$1#{placeholder}$2")
-      msg = msg.replace(new RegExp("([^a-zA-Z0-9_<]+)#{qm}$", "g"),      "$1#{placeholder}")
+    # Take the original message with no punctuation
+    if @master.unicodePunctuation?
+      pattern = msg.replace(@master.unicodePunctuation, "")
+    else
+      pattern = msg.replace(/[.,!?;:]/g, "")
 
-    # Convert the placeholders back in.
     tries = 0
-    while msg.indexOf("\x00") > -1
-      tries++
-      if tries > 50
-        @warn "Too many loops in substitution placeholders!"
+    giveup = 0
+    subgiveup = 0
+
+    # Look for words/phrases until there is no "spaces" in pattern
+    while pattern.indexOf(" ") > -1
+      giveup++
+      # Give up if there are too many substitutions (for safety)
+      if giveup >= 1000
+        @warn "Too many loops when handling substitutions!"
         break
 
-      match = msg.match("\\x00<(.+?)>\\x00")
-      if match
-        cap = parseInt(match[1])
-        result = ph[cap]
-        msg = msg.replace(new RegExp("\x00<#{cap}>\x00", "g"), result)
+      li = utils.nIndexOf(pattern, " ", maxwords)
+      subpattern = pattern.substring(0, li)
 
-    return msg
+      # If finds the pattern in sub object replace and stop to look
+      result = subs[subpattern];
+      if result!=undefined
+          msg = msg.replace(subpattern, result)
+      else
+        # Otherwise Look for substitutions in a subpattern
+        while subpattern.indexOf(" ") > -1
+            subgiveup++
+            # Give up if there are too many substitutions (for safety)
+            if subgiveup >= 1000
+                @warn("Too many loops when handling substitutions!")
+                break
+
+              li = subpattern.lastIndexOf(" ");
+              subpattern = subpattern.substring(0, li);
+
+              # If finds the subpattern in sub object replace and stop to look
+              result = subs[subpattern];
+              if result!=undefined
+                  msg = msg.replace(subpattern, result)
+                  break
+
+              tries++;
 
+      fi = pattern.indexOf(" ")
+      pattern = pattern.substring(fi+1)
+      tries++
+
+    # After all loops, see if just one word is in the pattern
+    result = subs[pattern]
+    if result!=undefined
+        msg = msg.replace(pattern, result)
+
+    return msg
 
 module.exports = Brain

src/rivescript.coffee

@@ -176,7 +176,9 @@ class RiveScript
     @_global   = {} # 'global' variables
     @_var      = {} # 'bot' variables
     @_sub      = {} # 'sub' substitutions
+    @_submax   = 1  # 'submax' max words in sub object
     @_person   = {} # 'person' substitutions
+    @_personmax= 1  # 'personmax' max words in person object
     @_array    = {} # 'array' variables
     @_users    = {} # 'user' variables
     @_freeze   = {} # frozen 'user' variables
@@ -463,6 +465,8 @@ class RiveScript
       continue unless ast.begin.hasOwnProperty type
       internal = "_#{type}" # so "global" maps to this._global
       for name, value of vars
+        if type=='sub' || type=='person'
+          @[internal+"max"] = Math.max(@[internal+"max"], name.split(" ").length);
         continue unless vars.hasOwnProperty name
         if value is "<undef>"
           delete @[internal][name]
@@ -706,6 +710,7 @@ class RiveScript
     if value is undefined
       delete @_sub[name]
     else
+      @_submax = Math.max(name.split(' ').length, @_submax)
       @_sub[name] = value
 
   ##
@@ -718,6 +723,7 @@ class RiveScript
     if value is undefined
       delete @_person[name]
     else
+      @_personmax = Math.max(name.split(' ').length, @_personmax)
       @_person[name] = value
 
   ##

src/utils.coffee

@@ -154,3 +154,19 @@ exports.isAPromise = (obj) ->
   typeof obj.then is 'function' and
   typeof obj.catch is 'function' and
   typeof obj.finally is 'function'
+
+##
+# int nIndexOf (string, string match, int index)
+#
+# Finds a match in a string at a given index
+#
+# Usage:
+# string = "My name is Rive"
+# match = " "
+# index = 2
+# return = 7
+#
+# Summary: It will look for a second space in the string
+##
+exports.nIndexOf = (string, match, index) ->
+   return string.split(match, index).join(match).length